Get started with Semwright

Connect your agent to supported applications on your computer.

Semwright connects your AI agent to supported applications on your computer. Your agent can discover available operations, request changes with your permission and inspect the result.

Semwright 1.0.0 is available in native bundles for Linux, macOS and Windows. See installation and setup for downloads and compatibility for the evidence and security-review status. Semwright is not a model or a hosted agent; you bring the agent.

Connect your agent

Follow Install and connect your agent to install Semwright, connect your existing agent and make a first read-only request. Then choose the application you want to use.

If you are contributing to Semwright, start with Build from source and First commands.

Understand the path

An agent discovers an operation and supplies structured inputs. The broker checks capability, policy, references and execution context before routing the request. Results carry execution provenance.

CLI / MCP / recipes / inspector
               ↓
      broker / policy / audit
               ↓
        provider runtime
               ↓
 native APIs / app drivers / MCP providers

Changing the frontend does not create a more privileged route. Discovery describes what a capability can do; it does not authorize its use.

Choose a reading path

Your next step	Reading path
Connect an agent	MCP, then capabilities and references
Integrate your application	Native SDK, then drivers and providers
Coordinate a project	Project Graph, artifact handoff and Effects
Reuse successful work	Skills, Recipes and workflow distillation
Inspect the boundary	Architecture, permissions and compatibility

The public website is static. Its examples never connect to a local broker or execute commands on your computer.

Source: architecture at the reviewed revision. Source references checked 7 October 2026.